add models

.gitignore

@@ -11,3 +11,5 @@ eval-results/
 eval-queue-bk/
 eval-results-bk/
 logs/
+
+.history

.pre-commit-config.yaml

@@ -1,53 +0,0 @@
-# Copyright (c) 2022, NVIDIA CORPORATION.  All rights reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-default_language_version:
-  python: python3
-
-ci:
-  autofix_prs: true
-  autoupdate_commit_msg: '[pre-commit.ci] pre-commit suggestions'
-  autoupdate_schedule: quarterly
-
-repos:
-  - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v4.3.0
-    hooks:
-      - id: check-yaml
-      - id: check-case-conflict
-      - id: detect-private-key
-      - id: check-added-large-files
-        args: ['--maxkb=1000']
-      - id: requirements-txt-fixer
-      - id: end-of-file-fixer
-      - id: trailing-whitespace
-
-  - repo: https://github.com/PyCQA/isort
-    rev: 5.12.0
-    hooks:
-      - id: isort
-        name: Format imports
-
-  - repo: https://github.com/psf/black
-    rev: 22.12.0
-    hooks:
-      - id: black
-        name: Format code
-        additional_dependencies: ['click==8.0.2']
-
-  - repo: https://github.com/charliermarsh/ruff-pre-commit
-    # Ruff version.
-    rev: 'v0.0.267'
-    hooks:
-      - id: ruff

Makefile

@@ -1,13 +0,0 @@
-.PHONY: style format
-
-
-style:
-	python -m black --line-length 119 .
-	python -m isort .
-	ruff check --fix .
-
-
-quality:
-	python -m black --check --line-length 119 .
-	python -m isort --check-only .
-	ruff check .

STRUCT_RES.json

@@ -0,0 +1,222 @@
+{
+    "deepseek-v2-lite-chat": {
+        "BaseARC": 69.01287553648069,
+        "BaseOpenbook": 64.60000000000001,
+        "📝StructARC": 40.155538694992416,
+        "📝StructOpenbook": 41.98437130002368,
+        "📝StructMMLU": 40.95659182440028,
+        "📝StructAverage": 41.03216727313879,
+        "BaseMMLU": 55.7,
+        "BaseAverage": 63.10429184549357
+    },
+    "baichuan2-7b-chat": {
+        "BaseARC": 57.596566523605155,
+        "BaseOpenbook": 60.8,
+        "📝StructARC": 31.39226100151745,
+        "📝StructOpenbook": 33.838503433578026,
+        "📝StructMMLU": 32.019719285283976,
+        "📝StructAverage": 32.41682790679315,
+        "BaseMMLU": 52.49,
+        "BaseAverage": 56.962188841201716
+    },
+    "mistral-7b-instruct-v0.3": {
+        "BaseARC": 76.39484978540773,
+        "BaseOpenbook": 74.4,
+        "📝StructARC": 42.640364188163886,
+        "📝StructOpenbook": 43.6182808430026,
+        "📝StructMMLU": 41.756990621302315,
+        "📝StructAverage": 42.67187855082293,
+        "BaseMMLU": 61.63,
+        "BaseAverage": 70.80828326180257
+    },
+    "yi-6b-chat": {
+        "BaseARC": 59.227467811158796,
+        "BaseOpenbook": 55.60000000000001,
+        "📝StructARC": 38.39150227617603,
+        "📝StructOpenbook": 38.977030547004496,
+        "📝StructMMLU": 36.34025031348594,
+        "📝StructAverage": 37.90292771222215,
+        "BaseMMLU": 58.24,
+        "BaseAverage": 57.68915593705293
+    },
+    "yi-1.5-9b-chat": {
+        "BaseARC": 85.75107296137338,
+        "BaseOpenbook": 83.0,
+        "📝StructARC": 57.01820940819423,
+        "📝StructOpenbook": 54.53469097797774,
+        "📝StructMMLU": 53.431308242228596,
+        "📝StructAverage": 54.99473620946685,
+        "BaseMMLU": 69.5,
+        "BaseAverage": 79.4170243204578
+    },
+    "llama-2-7b-chat": {
+        "BaseARC": 53.648068669527895,
+        "BaseOpenbook": 57.99999999999999,
+        "📝StructARC": 29.429059180576633,
+        "📝StructOpenbook": 33.24650722235378,
+        "📝StructMMLU": 27.884038239302765,
+        "📝StructAverage": 30.186534880744393,
+        "BaseMMLU": 34.1,
+        "BaseAverage": 48.5826895565093
+    },
+    "llama-3-8b-instruct": {
+        "BaseARC": 80.68669527896995,
+        "BaseOpenbook": 79.60000000000001,
+        "📝StructARC": 56.74317147192717,
+        "📝StructOpenbook": 57.06843476201753,
+        "📝StructMMLU": 54.63728961686302,
+        "📝StructAverage": 56.149631950269246,
+        "BaseMMLU": 68.4,
+        "BaseAverage": 76.22889842632333
+    },
+    "qwen1.5-7b-chat": {
+        "BaseARC": 70.21459227467811,
+        "BaseOpenbook": 68.60000000000001,
+        "📝StructARC": 40.92374810318665,
+        "📝StructOpenbook": 41.439734785697375,
+        "📝StructMMLU": 39.75951206275273,
+        "📝StructAverage": 40.707664983878914,
+        "BaseMMLU": 61.70508772,
+        "BaseAverage": 66.83989333155938
+    },
+    "qwen2-7b-instruct": {
+        "BaseARC": 84.29184549356223,
+        "BaseOpenbook": 82.39999999999999,
+        "📝StructARC": 58.34597875569044,
+        "📝StructOpenbook": 57.28155339805825,
+        "📝StructMMLU": 56.78738087582798,
+        "📝StructAverage": 57.47163767652555,
+        "BaseMMLU": 70.5,
+        "BaseAverage": 79.06394849785407
+    },
+    "mistral-7b-v0.3": {
+        "BaseARC": 76.65236051502146,
+        "BaseOpenbook": 72.8,
+        "📝StructARC": 46.936646433990894,
+        "📝StructOpenbook": 48.23585129055174,
+        "📝StructMMLU": 46.05517926584278,
+        "📝StructAverage": 47.075892330128475,
+        "BaseMMLU": 60.1,
+        "BaseAverage": 69.85078683834048
+    },
+    "yi-1.5-9b": {
+        "BaseARC": 88.41201716738198,
+        "BaseOpenbook": 83.0,
+        "📝StructARC": 60.60318664643399,
+        "📝StructOpenbook": 59.2943405162207,
+        "📝StructMMLU": 57.41526232409641,
+        "📝StructAverage": 59.10426316225036,
+        "BaseMMLU": 69.5,
+        "BaseAverage": 80.30400572246066
+    },
+    "yi-6b": {
+        "BaseARC": 77.6824034334764,
+        "BaseOpenbook": 73.6,
+        "📝StructARC": 46.936646433990894,
+        "📝StructOpenbook": 49.51456310679612,
+        "📝StructMMLU": 46.21888085492341,
+        "📝StructAverage": 47.55669679857014,
+        "BaseMMLU": 63.2,
+        "BaseAverage": 71.49413447782545
+    },
+    "deepseek-v2-lite": {
+        "BaseARC": 68.24034334763948,
+        "BaseOpenbook": 66.4,
+        "📝StructARC": 40.73406676783004,
+        "📝StructOpenbook": 42.97892493488042,
+        "📝StructMMLU": 41.024161265238256,
+        "📝StructAverage": 41.579050989316244,
+        "BaseMMLU": 58.3,
+        "BaseAverage": 64.3134477825465
+    },
+    "llama-3-8b": {
+        "BaseARC": 76.22317596566523,
+        "BaseOpenbook": 75.0,
+        "📝StructARC": 52.78831562974203,
+        "📝StructOpenbook": 53.04286052569264,
+        "📝StructMMLU": 51.37044261097523,
+        "📝StructAverage": 52.4005395888033,
+        "BaseMMLU": 66.6,
+        "BaseAverage": 72.60772532188841
+    },
+    "llama-2-7b": {
+        "BaseARC": 48.66952789699571,
+        "BaseOpenbook": 43.6,
+        "📝StructARC": 29.040212443095598,
+        "📝StructOpenbook": 31.730996921619703,
+        "📝StructMMLU": 29.187423497033315,
+        "📝StructAverage": 29.986210953916203,
+        "BaseMMLU": 45.7,
+        "BaseAverage": 45.98984263233191
+    },
+    "qwen2-7b": {
+        "BaseARC": 86.18025751072962,
+        "BaseOpenbook": 84.2,
+        "📝StructARC": 66.17033383915023,
+        "📝StructOpenbook": 66.13781671797301,
+        "📝StructMMLU": 65.24702121505192,
+        "📝StructAverage": 65.85172392405839,
+        "BaseMMLU": 70.3,
+        "BaseAverage": 80.22675250357655
+    },
+    "qwen1.5-7b": {
+        "BaseARC": 78.54077253218884,
+        "BaseOpenbook": 78.8,
+        "📝StructARC": 48.501517450682854,
+        "📝StructOpenbook": 50.248638408714186,
+        "📝StructMMLU": 46.709001366105916,
+        "📝StructAverage": 48.48638574183432,
+        "BaseMMLU": 61.0,
+        "BaseAverage": 72.78025751072961
+    },
+    "baichuan2-7b-base": {
+        "BaseARC": 61.630901287553655,
+        "BaseOpenbook": 62.6,
+        "📝StructARC": 32.42602427921092,
+        "📝StructOpenbook": 37.08264267108691,
+        "📝StructMMLU": 34.59571358681649,
+        "📝StructAverage": 34.7014601790381,
+        "BaseMMLU": 54.16,
+        "BaseAverage": 59.46363376251788
+    },
+    "llama-3-70b": {
+        "BaseARC": 92.1030042918455,
+        "BaseOpenbook": 89.60000000000001,
+        "📝StructARC": 73.12215477996965,
+        "📝StructOpenbook": 72.6024153445418,
+        "📝StructMMLU": 70.85211858440155,
+        "📝StructAverage": 72.19222956963766,
+        "BaseMMLU": 79.5,
+        "BaseAverage": 87.06766809728184
+    },
+    "qwen2-72b": {
+        "BaseARC": 95.79399141630901,
+        "BaseOpenbook": 96.39999999999999,
+        "📝StructARC": 79.10660091047042,
+        "📝StructOpenbook": 78.4986976083353,
+        "📝StructMMLU": 78.57972263190929,
+        "📝StructAverage": 78.72834038357168,
+        "BaseMMLU": 84.2,
+        "BaseAverage": 92.131330472103
+    },
+    "yi-1.5-34b": {
+        "BaseARC": 92.1030042918455,
+        "BaseOpenbook": 83.6,
+        "📝StructARC": 64.37784522003035,
+        "📝StructOpenbook": 63.76983187307601,
+        "📝StructMMLU": 63.296635991674755,
+        "📝StructAverage": 63.81477102826037,
+        "BaseMMLU": 77.1,
+        "BaseAverage": 84.26766809728183
+    },
+    "mixtral-8x7b-v0.1": {
+        "BaseARC": 84.0343347639485,
+        "BaseOpenbook": 77.8,
+        "📝StructARC": 62.3103186646434,
+        "📝StructOpenbook": 62.32536111768885,
+        "📝StructMMLU": 60.46643846946602,
+        "📝StructAverage": 61.70070608393275,
+        "BaseMMLU": 70.6,
+        "BaseAverage": 77.47811158798282
+    }
+}

app.py

@@ -1,204 +1,137 @@
+import json
+
 import gradio as gr
-from gradio_leaderboard import Leaderboard, ColumnFilter, SelectColumns
 import pandas as pd
-from apscheduler.schedulers.background import BackgroundScheduler
-from huggingface_hub import snapshot_download
 
-from src.about import (
-    CITATION_BUTTON_LABEL,
-    CITATION_BUTTON_TEXT,
-    EVALUATION_QUEUE_TEXT,
-    INTRODUCTION_TEXT,
-    LLM_BENCHMARKS_TEXT,
-    TITLE,
-)
-from src.display.css_html_js import custom_css
-from src.display.utils import (
-    BENCHMARK_COLS,
-    COLS,
-    EVAL_COLS,
-    EVAL_TYPES,
+from css_html import custom_css
+from text_content import ABOUT_TEXT, CITATION_BUTTON_TEXT, CITATION_BUTTON_LABEL, ACKNOWLEDGEMENT_TEXT, NOTES_TEXT, HEAD_TEXT
+from utils import (
     AutoEvalColumn,
-    ModelType,
     fields,
-    WeightType,
-    Precision
 )
-from src.envs import API, EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, QUEUE_REPO, REPO_ID, RESULTS_REPO, TOKEN
-from src.populate import get_evaluation_queue_df, get_leaderboard_df
-from src.submission.submit import add_new_eval
-
-
-def restart_space():
-    API.restart_space(repo_id=REPO_ID)
-
-### Space initialisation
-try:
-    print(EVAL_REQUESTS_PATH)
-    snapshot_download(
-        repo_id=QUEUE_REPO, local_dir=EVAL_REQUESTS_PATH, repo_type="dataset", tqdm_class=None, etag_timeout=30, token=TOKEN
-    )
-except Exception:
-    restart_space()
-try:
-    print(EVAL_RESULTS_PATH)
-    snapshot_download(
-        repo_id=RESULTS_REPO, local_dir=EVAL_RESULTS_PATH, repo_type="dataset", tqdm_class=None, etag_timeout=30, token=TOKEN
-    )
-except Exception:
-    restart_space()
-
-
-LEADERBOARD_DF = get_leaderboard_df(EVAL_RESULTS_PATH, EVAL_REQUESTS_PATH, COLS, BENCHMARK_COLS)
-
-(
-    finished_eval_queue_df,
-    running_eval_queue_df,
-    pending_eval_queue_df,
-) = get_evaluation_queue_df(EVAL_REQUESTS_PATH, EVAL_COLS)
-
-def init_leaderboard(dataframe):
-    if dataframe is None or dataframe.empty:
-        raise ValueError("Leaderboard DataFrame is empty or None.")
-    return Leaderboard(
-        value=dataframe,
-        datatype=[c.type for c in fields(AutoEvalColumn)],
-        select_columns=SelectColumns(
-            default_selection=[c.name for c in fields(AutoEvalColumn) if c.displayed_by_default],
-            cant_deselect=[c.name for c in fields(AutoEvalColumn) if c.never_hidden],
-            label="Select Columns to Display:",
-        ),
-        search_columns=[AutoEvalColumn.model.name, AutoEvalColumn.license.name],
-        hide_columns=[c.name for c in fields(AutoEvalColumn) if c.hidden],
-        filter_columns=[
-            ColumnFilter(AutoEvalColumn.model_type.name, type="checkboxgroup", label="Model types"),
-            ColumnFilter(AutoEvalColumn.precision.name, type="checkboxgroup", label="Precision"),
-            ColumnFilter(
-                AutoEvalColumn.params.name,
-                type="slider",
-                min=0.01,
-                max=150,
-                label="Select the number of parameters (B)",
-            ),
-            ColumnFilter(
-                AutoEvalColumn.still_on_hub.name, type="boolean", label="Deleted/incomplete", default=True
-            ),
-        ],
-        bool_checkboxgroup_label="Hide models",
-        interactive=False,
-    )
 
+result_path = './STRUCT_RES.json'
+with open(result_path, 'r') as f:
+    data = json.load(f)
+    
+    rows = []
+    for model in data:
+        row = {"model": model}
+        for key in data[model]:
+            print(key)
+            row[key] = round(data[model][key], 2)
+        rows.append(row)
+        
+    df = pd.DataFrame(rows)
+    df = df.sort_values(by='📝StructAverage', ascending=False)
+
+COLS = [c.name for c in fields(AutoEvalColumn) if not c.hidden]
+TYPES = [c.type for c in fields(AutoEvalColumn) if not c.hidden]
+
+COLS_LITE = [
+    c.name for c in fields(AutoEvalColumn) if c.displayed_by_default and not c.hidden
+]
+
+TYPES_LITE = [
+    c.type for c in fields(AutoEvalColumn) if c.displayed_by_default and not c.hidden
+]
+
+def select_columns(df, columns):
+    always_here_cols = [
+        AutoEvalColumn.model.name,
+    ]
+    # We use COLS to maintain sorting
+    filtered_df = df[
+        always_here_cols + [c for c in COLS if c in df.columns and c in columns]
+    ]
+    return filtered_df
 
 demo = gr.Blocks(css=custom_css)
 with demo:
-    gr.HTML(TITLE)
-    gr.Markdown(INTRODUCTION_TEXT, elem_classes="markdown-text")
+    with gr.Column():
+        gr.Markdown(
+            """<div style="text-align: center;"><h1> 🏅 StructEval Leaderboard</h1></div>\
+            <br>\
+""",
+            elem_classes="markdown-text",
+        )
+        
+        gr.Markdown(HEAD_TEXT, elem_classes="markdown-text")
 
     with gr.Tabs(elem_classes="tab-buttons") as tabs:
-        with gr.TabItem("🏅 LLM Benchmark", elem_id="llm-benchmark-tab-table", id=0):
-            leaderboard = init_leaderboard(LEADERBOARD_DF)
-
-        with gr.TabItem("📝 About", elem_id="llm-benchmark-tab-table", id=2):
-            gr.Markdown(LLM_BENCHMARKS_TEXT, elem_classes="markdown-text")
-
-        with gr.TabItem("🚀 Submit here! ", elem_id="llm-benchmark-tab-table", id=3):
-            with gr.Column():
-                with gr.Row():
-                    gr.Markdown(EVALUATION_QUEUE_TEXT, elem_classes="markdown-text")
-
-                with gr.Column():
-                    with gr.Accordion(
-                        f"✅ Finished Evaluations ({len(finished_eval_queue_df)})",
-                        open=False,
-                    ):
-                        with gr.Row():
-                            finished_eval_table = gr.components.Dataframe(
-                                value=finished_eval_queue_df,
-                                headers=EVAL_COLS,
-                                datatype=EVAL_TYPES,
-                                row_count=5,
-                            )
-                    with gr.Accordion(
-                        f"🔄 Running Evaluation Queue ({len(running_eval_queue_df)})",
-                        open=False,
-                    ):
-                        with gr.Row():
-                            running_eval_table = gr.components.Dataframe(
-                                value=running_eval_queue_df,
-                                headers=EVAL_COLS,
-                                datatype=EVAL_TYPES,
-                                row_count=5,
-                            )
-
-                    with gr.Accordion(
-                        f"⏳ Pending Evaluation Queue ({len(pending_eval_queue_df)})",
-                        open=False,
-                    ):
-                        with gr.Row():
-                            pending_eval_table = gr.components.Dataframe(
-                                value=pending_eval_queue_df,
-                                headers=EVAL_COLS,
-                                datatype=EVAL_TYPES,
-                                row_count=5,
+        with gr.Column():
+            with gr.Tabs(elem_classes="A100-tabs") as A100_tabs:
+                with gr.TabItem("🔍 Evaluation Table", id=0):
+                    with gr.Column():
+                        with gr.Accordion("⏬ Hidden Columns", open=False):
+                            shown_columns = gr.CheckboxGroup(
+                                choices=[
+                                    c
+                                    for c in COLS
+                                    if c
+                                    not in [
+                                        AutoEvalColumn.model.name,
+                                    ]
+                                ],
+                                value=[
+                                    c
+                                    for c in COLS_LITE
+                                    if c
+                                    not in [
+                                        AutoEvalColumn.model.name,
+                                    ]
+                                ],
+                                label="",
+                                elem_id="column-select",
+                                interactive=True,
                             )
-            with gr.Row():
-                gr.Markdown("# ✉️✨ Submit your model here!", elem_classes="markdown-text")
-
-            with gr.Row():
-                with gr.Column():
-                    model_name_textbox = gr.Textbox(label="Model name")
-                    revision_name_textbox = gr.Textbox(label="Revision commit", placeholder="main")
-                    model_type = gr.Dropdown(
-                        choices=[t.to_str(" : ") for t in ModelType if t != ModelType.Unknown],
-                        label="Model type",
-                        multiselect=False,
-                        value=None,
-                        interactive=True,
+                            
+                    leaderboard_df = gr.components.Dataframe(
+                        value=df[
+                            [
+                                AutoEvalColumn.model.name,
+                            ]
+                            + shown_columns.value
+                        ],
+                        headers=COLS,
+                        datatype=TYPES,
+                        elem_id="leaderboard-table",
+                        interactive=False,
                     )
 
-                with gr.Column():
-                    precision = gr.Dropdown(
-                        choices=[i.value.name for i in Precision if i != Precision.Unknown],
-                        label="Precision",
-                        multiselect=False,
-                        value="float16",
-                        interactive=True,
+                    hidden_leaderboard_df = gr.components.Dataframe(
+                        value=df,
+                        headers=COLS,
+                        datatype=["str" for _ in range(len(COLS))],
+                        visible=False,
                     )
-                    weight_type = gr.Dropdown(
-                        choices=[i.value.name for i in WeightType],
-                        label="Weights type",
-                        multiselect=False,
-                        value="Original",
-                        interactive=True,
+                    
+                    shown_columns.change(
+                        select_columns,
+                        [hidden_leaderboard_df, shown_columns],
+                        leaderboard_df,
                     )
-                    base_model_name_textbox = gr.Textbox(label="Base model (for delta or adapter weights)")
-
-            submit_button = gr.Button("Submit Eval")
-            submission_result = gr.Markdown()
-            submit_button.click(
-                add_new_eval,
-                [
-                    model_name_textbox,
-                    base_model_name_textbox,
-                    revision_name_textbox,
-                    precision,
-                    weight_type,
-                    model_type,
-                ],
-                submission_result,
-            )
-
-    with gr.Row():
-        with gr.Accordion("📙 Citation", open=False):
-            citation_button = gr.Textbox(
-                value=CITATION_BUTTON_TEXT,
-                label=CITATION_BUTTON_LABEL,
-                lines=20,
-                elem_id="citation-button",
-                show_copy_button=True,
-            )
-
-scheduler = BackgroundScheduler()
-scheduler.add_job(restart_space, "interval", seconds=1800)
-scheduler.start()
-demo.queue(default_concurrency_limit=40).launch()
+                
+                        
+                with gr.TabItem("📝 About", id=1):
+                    gr.Markdown(ABOUT_TEXT, elem_classes="markdown-text")
+        
+        with gr.Row():
+            with gr.Accordion("📒 Notes"):
+                gr.Markdown(NOTES_TEXT, elem_classes="markdown-text")
+        
+        with gr.Row():
+            with gr.Accordion("📜 Citation", open=False):
+                citation_button = gr.Textbox(
+                    value=CITATION_BUTTON_TEXT,
+                    label=CITATION_BUTTON_LABEL,
+                    lines=10,
+                    elem_id="citation-button",
+                    show_copy_button=True,
+                )
+        
+        with gr.Row():
+            with gr.Accordion("🙏 Acknowledgement", open=False):
+                gr.Markdown(ACKNOWLEDGEMENT_TEXT)
+
+demo.launch()

src/display/css_html_js.py → css_html.py

@@ -1,34 +1,32 @@
 custom_css = """
-
+#changelog-text {
+    font-size: 16px !important;
+}
+#changelog-text h2 {
+    font-size: 18px !important;
+}
 .markdown-text {
     font-size: 16px !important;
 }
-
 #models-to-add-text {
     font-size: 18px !important;
 }
-
 #citation-button span {
     font-size: 16px !important;
 }
-
 #citation-button textarea {
     font-size: 16px !important;
 }
-
 #citation-button > label > button {
     margin: 6px;
     transform: scale(1.3);
 }
-
 #leaderboard-table {
     margin-top: 15px
 }
-
 #leaderboard-table-lite {
     margin-top: 15px
 }
-
 #search-bar-table-box > div:first-child {
     background: none;
     border: none;
@@ -37,7 +35,11 @@ custom_css = """
 #search-bar {
     padding: 0px;
 }
-
+/* Hides the final AutoEvalColumn */
+#llm-benchmark-tab-table table td:last-child,
+#llm-benchmark-tab-table table th:last-child {
+    display: none;
+}
 /* Limit the width of the first AutoEvalColumn so that names don't expand too much */
 table td:first-child,
 table th:first-child {
@@ -45,11 +47,9 @@ table th:first-child {
     overflow: auto;
     white-space: nowrap;
 }
-
 .tab-buttons button {
     font-size: 20px;
 }
-
 #scale-logo {
     border-style: none !important;
     box-shadow: none;
@@ -58,48 +58,7 @@ table th:first-child {
     margin-right: auto;
     max-width: 600px;
 }
-
 #scale-logo .download {
     display: none;
 }
-#filter_type{
-    border: 0;
-    padding-left: 0;
-    padding-top: 0;
-}
-#filter_type label {
-    display: flex;
-}
-#filter_type label > span{
-    margin-top: var(--spacing-lg);
-    margin-right: 0.5em;
-}
-#filter_type label > .wrap{
-    width: 103px;
-}
-#filter_type label > .wrap .wrap-inner{  
-    padding: 2px;
-}
-#filter_type label > .wrap .wrap-inner input{
-    width: 1px
-}
-#filter-columns-type{
-    border:0;
-    padding:0.5;
-}
-#filter-columns-size{
-    border:0;
-    padding:0.5;
-}
-#box-filter > .form{
-    border: 0
-}
-"""
-
-get_window_url_params = """
-    function(url_params) {
-        const params = new URLSearchParams(window.location.search);
-        url_params = Object.fromEntries(params);
-        return url_params;
-    }
-    """
+"""

pyproject.toml

@@ -1,13 +0,0 @@
-[tool.ruff]
-# Enable pycodestyle (`E`) and Pyflakes (`F`) codes by default.
-select = ["E", "F"]
-ignore = ["E501"] # line too long (black is taking care of this)
-line-length = 119
-fixable = ["A", "B", "C", "D", "E", "F", "G", "I", "N", "Q", "S", "T", "W", "ANN", "ARG", "BLE", "COM", "DJ", "DTZ", "EM", "ERA", "EXE", "FBT", "ICN", "INP", "ISC", "NPY", "PD", "PGH", "PIE", "PL", "PT", "PTH", "PYI", "RET", "RSE", "RUF", "SIM", "SLF", "TCH", "TID", "TRY", "UP", "YTT"]
-
-[tool.isort]
-profile = "black"
-line_length = 119
-
-[tool.black]
-line-length = 119

src/about.py

@@ -1,72 +0,0 @@
-from dataclasses import dataclass
-from enum import Enum
-
-@dataclass
-class Task:
-    benchmark: str
-    metric: str
-    col_name: str
-
-
-# Select your tasks here
-# ---------------------------------------------------
-class Tasks(Enum):
-    # task_key in the json file, metric_key in the json file, name to display in the leaderboard 
-    task0 = Task("anli_r1", "acc", "ANLI")
-    task1 = Task("logiqa", "acc_norm", "LogiQA")
-
-NUM_FEWSHOT = 0 # Change with your few shot
-# ---------------------------------------------------
-
-
-
-# Your leaderboard name
-TITLE = """<h1 align="center" id="space-title">Demo leaderboard</h1>"""
-
-# What does your leaderboard evaluate?
-INTRODUCTION_TEXT = """
-Intro text
-"""
-
-# Which evaluations are you running? how can people reproduce what you have?
-LLM_BENCHMARKS_TEXT = f"""
-## How it works
-
-## Reproducibility
-To reproduce our results, here is the commands you can run:
-
-"""
-
-EVALUATION_QUEUE_TEXT = """
-## Some good practices before submitting a model
-
-### 1) Make sure you can load your model and tokenizer using AutoClasses:
-```python
-from transformers import AutoConfig, AutoModel, AutoTokenizer
-config = AutoConfig.from_pretrained("your model name", revision=revision)
-model = AutoModel.from_pretrained("your model name", revision=revision)
-tokenizer = AutoTokenizer.from_pretrained("your model name", revision=revision)
-```
-If this step fails, follow the error messages to debug your model before submitting it. It's likely your model has been improperly uploaded.
-
-Note: make sure your model is public!
-Note: if your model needs `use_remote_code=True`, we do not support this option yet but we are working on adding it, stay posted!
-
-### 2) Convert your model weights to [safetensors](https://huggingface.co/docs/safetensors/index)
-It's a new format for storing weights which is safer and faster to load and use. It will also allow us to add the number of parameters of your model to the `Extended Viewer`!
-
-### 3) Make sure your model has an open license!
-This is a leaderboard for Open LLMs, and we'd love for as many people as possible to know they can use your model 🤗
-
-### 4) Fill up your model card
-When we add extra information about models to the leaderboard, it will be automatically taken from the model card
-
-## In case of model failure
-If your model is displayed in the `FAILED` category, its execution stopped.
-Make sure you have followed the above steps first.
-If everything is done, check you can launch the EleutherAIHarness on your model locally, using the above command without modifications (you can add `--limit` to limit the number of examples per task).
-"""
-
-CITATION_BUTTON_LABEL = "Copy the following snippet to cite these results"
-CITATION_BUTTON_TEXT = r"""
-"""

src/display/formatting.py

@@ -1,27 +0,0 @@
-def model_hyperlink(link, model_name):
-    return f'<a target="_blank" href="{link}" style="color: var(--link-text-color); text-decoration: underline;text-decoration-style: dotted;">{model_name}</a>'
-
-
-def make_clickable_model(model_name):
-    link = f"https://huggingface.co/{model_name}"
-    return model_hyperlink(link, model_name)
-
-
-def styled_error(error):
-    return f"<p style='color: red; font-size: 20px; text-align: center;'>{error}</p>"
-
-
-def styled_warning(warn):
-    return f"<p style='color: orange; font-size: 20px; text-align: center;'>{warn}</p>"
-
-
-def styled_message(message):
-    return f"<p style='color: green; font-size: 20px; text-align: center;'>{message}</p>"
-
-
-def has_no_nan_values(df, columns):
-    return df[columns].notna().all(axis=1)
-
-
-def has_nan_values(df, columns):
-    return df[columns].isna().any(axis=1)

src/display/utils.py

@@ -1,110 +0,0 @@
-from dataclasses import dataclass, make_dataclass
-from enum import Enum
-
-import pandas as pd
-
-from src.about import Tasks
-
-def fields(raw_class):
-    return [v for k, v in raw_class.__dict__.items() if k[:2] != "__" and k[-2:] != "__"]
-
-
-# These classes are for user facing column names,
-# to avoid having to change them all around the code
-# when a modif is needed
-@dataclass
-class ColumnContent:
-    name: str
-    type: str
-    displayed_by_default: bool
-    hidden: bool = False
-    never_hidden: bool = False
-
-## Leaderboard columns
-auto_eval_column_dict = []
-# Init
-auto_eval_column_dict.append(["model_type_symbol", ColumnContent, ColumnContent("T", "str", True, never_hidden=True)])
-auto_eval_column_dict.append(["model", ColumnContent, ColumnContent("Model", "markdown", True, never_hidden=True)])
-#Scores
-auto_eval_column_dict.append(["average", ColumnContent, ColumnContent("Average ⬆️", "number", True)])
-for task in Tasks:
-    auto_eval_column_dict.append([task.name, ColumnContent, ColumnContent(task.value.col_name, "number", True)])
-# Model information
-auto_eval_column_dict.append(["model_type", ColumnContent, ColumnContent("Type", "str", False)])
-auto_eval_column_dict.append(["architecture", ColumnContent, ColumnContent("Architecture", "str", False)])
-auto_eval_column_dict.append(["weight_type", ColumnContent, ColumnContent("Weight type", "str", False, True)])
-auto_eval_column_dict.append(["precision", ColumnContent, ColumnContent("Precision", "str", False)])
-auto_eval_column_dict.append(["license", ColumnContent, ColumnContent("Hub License", "str", False)])
-auto_eval_column_dict.append(["params", ColumnContent, ColumnContent("#Params (B)", "number", False)])
-auto_eval_column_dict.append(["likes", ColumnContent, ColumnContent("Hub ❤️", "number", False)])
-auto_eval_column_dict.append(["still_on_hub", ColumnContent, ColumnContent("Available on the hub", "bool", False)])
-auto_eval_column_dict.append(["revision", ColumnContent, ColumnContent("Model sha", "str", False, False)])
-
-# We use make dataclass to dynamically fill the scores from Tasks
-AutoEvalColumn = make_dataclass("AutoEvalColumn", auto_eval_column_dict, frozen=True)
-
-## For the queue columns in the submission tab
-@dataclass(frozen=True)
-class EvalQueueColumn:  # Queue column
-    model = ColumnContent("model", "markdown", True)
-    revision = ColumnContent("revision", "str", True)
-    private = ColumnContent("private", "bool", True)
-    precision = ColumnContent("precision", "str", True)
-    weight_type = ColumnContent("weight_type", "str", "Original")
-    status = ColumnContent("status", "str", True)
-
-## All the model information that we might need
-@dataclass
-class ModelDetails:
-    name: str
-    display_name: str = ""
-    symbol: str = "" # emoji
-
-
-class ModelType(Enum):
-    PT = ModelDetails(name="pretrained", symbol="🟢")
-    FT = ModelDetails(name="fine-tuned", symbol="🔶")
-    IFT = ModelDetails(name="instruction-tuned", symbol="⭕")
-    RL = ModelDetails(name="RL-tuned", symbol="🟦")
-    Unknown = ModelDetails(name="", symbol="?")
-
-    def to_str(self, separator=" "):
-        return f"{self.value.symbol}{separator}{self.value.name}"
-
-    @staticmethod
-    def from_str(type):
-        if "fine-tuned" in type or "🔶" in type:
-            return ModelType.FT
-        if "pretrained" in type or "🟢" in type:
-            return ModelType.PT
-        if "RL-tuned" in type or "🟦" in type:
-            return ModelType.RL
-        if "instruction-tuned" in type or "⭕" in type:
-            return ModelType.IFT
-        return ModelType.Unknown
-
-class WeightType(Enum):
-    Adapter = ModelDetails("Adapter")
-    Original = ModelDetails("Original")
-    Delta = ModelDetails("Delta")
-
-class Precision(Enum):
-    float16 = ModelDetails("float16")
-    bfloat16 = ModelDetails("bfloat16")
-    Unknown = ModelDetails("?")
-
-    def from_str(precision):
-        if precision in ["torch.float16", "float16"]:
-            return Precision.float16
-        if precision in ["torch.bfloat16", "bfloat16"]:
-            return Precision.bfloat16
-        return Precision.Unknown
-
-# Column selection
-COLS = [c.name for c in fields(AutoEvalColumn) if not c.hidden]
-
-EVAL_COLS = [c.name for c in fields(EvalQueueColumn)]
-EVAL_TYPES = [c.type for c in fields(EvalQueueColumn)]
-
-BENCHMARK_COLS = [t.value.col_name for t in Tasks]
-

src/envs.py

@@ -1,25 +0,0 @@
-import os
-
-from huggingface_hub import HfApi
-
-# Info to change for your repository
-# ----------------------------------
-TOKEN = os.environ.get("HF_TOKEN") # A read/write token for your org
-
-OWNER = "demo-leaderboard-backend" # Change to your org - don't forget to create a results and request dataset, with the correct format!
-# ----------------------------------
-
-REPO_ID = f"{OWNER}/leaderboard"
-QUEUE_REPO = f"{OWNER}/requests"
-RESULTS_REPO = f"{OWNER}/results"
-
-# If you setup a cache later, just change HF_HOME
-CACHE_PATH=os.getenv("HF_HOME", ".")
-
-# Local caches
-EVAL_REQUESTS_PATH = os.path.join(CACHE_PATH, "eval-queue")
-EVAL_RESULTS_PATH = os.path.join(CACHE_PATH, "eval-results")
-EVAL_REQUESTS_PATH_BACKEND = os.path.join(CACHE_PATH, "eval-queue-bk")
-EVAL_RESULTS_PATH_BACKEND = os.path.join(CACHE_PATH, "eval-results-bk")
-
-API = HfApi(token=TOKEN)

src/leaderboard/read_evals.py

@@ -1,196 +0,0 @@
-import glob
-import json
-import math
-import os
-from dataclasses import dataclass
-
-import dateutil
-import numpy as np
-
-from src.display.formatting import make_clickable_model
-from src.display.utils import AutoEvalColumn, ModelType, Tasks, Precision, WeightType
-from src.submission.check_validity import is_model_on_hub
-
-
-@dataclass
-class EvalResult:
-    """Represents one full evaluation. Built from a combination of the result and request file for a given run.
-    """
-    eval_name: str # org_model_precision (uid)
-    full_model: str # org/model (path on hub)
-    org: str 
-    model: str
-    revision: str # commit hash, "" if main
-    results: dict
-    precision: Precision = Precision.Unknown
-    model_type: ModelType = ModelType.Unknown # Pretrained, fine tuned, ...
-    weight_type: WeightType = WeightType.Original # Original or Adapter
-    architecture: str = "Unknown" 
-    license: str = "?"
-    likes: int = 0
-    num_params: int = 0
-    date: str = "" # submission date of request file
-    still_on_hub: bool = False
-
-    @classmethod
-    def init_from_json_file(self, json_filepath):
-        """Inits the result from the specific model result file"""
-        with open(json_filepath) as fp:
-            data = json.load(fp)
-
-        config = data.get("config")
-
-        # Precision
-        precision = Precision.from_str(config.get("model_dtype"))
-
-        # Get model and org
-        org_and_model = config.get("model_name", config.get("model_args", None))
-        org_and_model = org_and_model.split("/", 1)
-
-        if len(org_and_model) == 1:
-            org = None
-            model = org_and_model[0]
-            result_key = f"{model}_{precision.value.name}"
-        else:
-            org = org_and_model[0]
-            model = org_and_model[1]
-            result_key = f"{org}_{model}_{precision.value.name}"
-        full_model = "/".join(org_and_model)
-
-        still_on_hub, _, model_config = is_model_on_hub(
-            full_model, config.get("model_sha", "main"), trust_remote_code=True, test_tokenizer=False
-        )
-        architecture = "?"
-        if model_config is not None:
-            architectures = getattr(model_config, "architectures", None)
-            if architectures:
-                architecture = ";".join(architectures)
-
-        # Extract results available in this file (some results are split in several files)
-        results = {}
-        for task in Tasks:
-            task = task.value
-
-            # We average all scores of a given metric (not all metrics are present in all files)
-            accs = np.array([v.get(task.metric, None) for k, v in data["results"].items() if task.benchmark == k])
-            if accs.size == 0 or any([acc is None for acc in accs]):
-                continue
-
-            mean_acc = np.mean(accs) * 100.0
-            results[task.benchmark] = mean_acc
-
-        return self(
-            eval_name=result_key,
-            full_model=full_model,
-            org=org,
-            model=model,
-            results=results,
-            precision=precision,  
-            revision= config.get("model_sha", ""),
-            still_on_hub=still_on_hub,
-            architecture=architecture
-        )
-
-    def update_with_request_file(self, requests_path):
-        """Finds the relevant request file for the current model and updates info with it"""
-        request_file = get_request_file_for_model(requests_path, self.full_model, self.precision.value.name)
-
-        try:
-            with open(request_file, "r") as f:
-                request = json.load(f)
-            self.model_type = ModelType.from_str(request.get("model_type", ""))
-            self.weight_type = WeightType[request.get("weight_type", "Original")]
-            self.license = request.get("license", "?")
-            self.likes = request.get("likes", 0)
-            self.num_params = request.get("params", 0)
-            self.date = request.get("submitted_time", "")
-        except Exception:
-            print(f"Could not find request file for {self.org}/{self.model} with precision {self.precision.value.name}")
-
-    def to_dict(self):
-        """Converts the Eval Result to a dict compatible with our dataframe display"""
-        average = sum([v for v in self.results.values() if v is not None]) / len(Tasks)
-        data_dict = {
-            "eval_name": self.eval_name,  # not a column, just a save name,
-            AutoEvalColumn.precision.name: self.precision.value.name,
-            AutoEvalColumn.model_type.name: self.model_type.value.name,
-            AutoEvalColumn.model_type_symbol.name: self.model_type.value.symbol,
-            AutoEvalColumn.weight_type.name: self.weight_type.value.name,
-            AutoEvalColumn.architecture.name: self.architecture,
-            AutoEvalColumn.model.name: make_clickable_model(self.full_model),
-            AutoEvalColumn.revision.name: self.revision,
-            AutoEvalColumn.average.name: average,
-            AutoEvalColumn.license.name: self.license,
-            AutoEvalColumn.likes.name: self.likes,
-            AutoEvalColumn.params.name: self.num_params,
-            AutoEvalColumn.still_on_hub.name: self.still_on_hub,
-        }
-
-        for task in Tasks:
-            data_dict[task.value.col_name] = self.results[task.value.benchmark]
-
-        return data_dict
-
-
-def get_request_file_for_model(requests_path, model_name, precision):
-    """Selects the correct request file for a given model. Only keeps runs tagged as FINISHED"""
-    request_files = os.path.join(
-        requests_path,
-        f"{model_name}_eval_request_*.json",
-    )
-    request_files = glob.glob(request_files)
-
-    # Select correct request file (precision)
-    request_file = ""
-    request_files = sorted(request_files, reverse=True)
-    for tmp_request_file in request_files:
-        with open(tmp_request_file, "r") as f:
-            req_content = json.load(f)
-            if (
-                req_content["status"] in ["FINISHED"]
-                and req_content["precision"] == precision.split(".")[-1]
-            ):
-                request_file = tmp_request_file
-    return request_file
-
-
-def get_raw_eval_results(results_path: str, requests_path: str) -> list[EvalResult]:
-    """From the path of the results folder root, extract all needed info for results"""
-    model_result_filepaths = []
-
-    for root, _, files in os.walk(results_path):
-        # We should only have json files in model results
-        if len(files) == 0 or any([not f.endswith(".json") for f in files]):
-            continue
-
-        # Sort the files by date
-        try:
-            files.sort(key=lambda x: x.removesuffix(".json").removeprefix("results_")[:-7])
-        except dateutil.parser._parser.ParserError:
-            files = [files[-1]]
-
-        for file in files:
-            model_result_filepaths.append(os.path.join(root, file))
-
-    eval_results = {}
-    for model_result_filepath in model_result_filepaths:
-        # Creation of result
-        eval_result = EvalResult.init_from_json_file(model_result_filepath)
-        eval_result.update_with_request_file(requests_path)
-
-        # Store results of same eval together
-        eval_name = eval_result.eval_name
-        if eval_name in eval_results.keys():
-            eval_results[eval_name].results.update({k: v for k, v in eval_result.results.items() if v is not None})
-        else:
-            eval_results[eval_name] = eval_result
-
-    results = []
-    for v in eval_results.values():
-        try:
-            v.to_dict() # we test if the dict version is complete
-            results.append(v)
-        except KeyError:  # not all eval values present
-            continue
-
-    return results

src/populate.py

@@ -1,58 +0,0 @@
-import json
-import os
-
-import pandas as pd
-
-from src.display.formatting import has_no_nan_values, make_clickable_model
-from src.display.utils import AutoEvalColumn, EvalQueueColumn
-from src.leaderboard.read_evals import get_raw_eval_results
-
-
-def get_leaderboard_df(results_path: str, requests_path: str, cols: list, benchmark_cols: list) -> pd.DataFrame:
-    """Creates a dataframe from all the individual experiment results"""
-    raw_data = get_raw_eval_results(results_path, requests_path)
-    all_data_json = [v.to_dict() for v in raw_data]
-
-    df = pd.DataFrame.from_records(all_data_json)
-    df = df.sort_values(by=[AutoEvalColumn.average.name], ascending=False)
-    df = df[cols].round(decimals=2)
-
-    # filter out if any of the benchmarks have not been produced
-    df = df[has_no_nan_values(df, benchmark_cols)]
-    return df
-
-
-def get_evaluation_queue_df(save_path: str, cols: list) -> list[pd.DataFrame]:
-    """Creates the different dataframes for the evaluation queues requestes"""
-    entries = [entry for entry in os.listdir(save_path) if not entry.startswith(".")]
-    all_evals = []
-
-    for entry in entries:
-        if ".json" in entry:
-            file_path = os.path.join(save_path, entry)
-            with open(file_path) as fp:
-                data = json.load(fp)
-
-            data[EvalQueueColumn.model.name] = make_clickable_model(data["model"])
-            data[EvalQueueColumn.revision.name] = data.get("revision", "main")
-
-            all_evals.append(data)
-        elif ".md" not in entry:
-            # this is a folder
-            sub_entries = [e for e in os.listdir(f"{save_path}/{entry}") if not e.startswith(".")]
-            for sub_entry in sub_entries:
-                file_path = os.path.join(save_path, entry, sub_entry)
-                with open(file_path) as fp:
-                    data = json.load(fp)
-
-                data[EvalQueueColumn.model.name] = make_clickable_model(data["model"])
-                data[EvalQueueColumn.revision.name] = data.get("revision", "main")
-                all_evals.append(data)
-
-    pending_list = [e for e in all_evals if e["status"] in ["PENDING", "RERUN"]]
-    running_list = [e for e in all_evals if e["status"] == "RUNNING"]
-    finished_list = [e for e in all_evals if e["status"].startswith("FINISHED") or e["status"] == "PENDING_NEW_EVAL"]
-    df_pending = pd.DataFrame.from_records(pending_list, columns=cols)
-    df_running = pd.DataFrame.from_records(running_list, columns=cols)
-    df_finished = pd.DataFrame.from_records(finished_list, columns=cols)
-    return df_finished[cols], df_running[cols], df_pending[cols]

src/submission/check_validity.py

@@ -1,99 +0,0 @@
-import json
-import os
-import re
-from collections import defaultdict
-from datetime import datetime, timedelta, timezone
-
-import huggingface_hub
-from huggingface_hub import ModelCard
-from huggingface_hub.hf_api import ModelInfo
-from transformers import AutoConfig
-from transformers.models.auto.tokenization_auto import AutoTokenizer
-
-def check_model_card(repo_id: str) -> tuple[bool, str]:
-    """Checks if the model card and license exist and have been filled"""
-    try:
-        card = ModelCard.load(repo_id)
-    except huggingface_hub.utils.EntryNotFoundError:
-        return False, "Please add a model card to your model to explain how you trained/fine-tuned it."
-
-    # Enforce license metadata
-    if card.data.license is None:
-        if not ("license_name" in card.data and "license_link" in card.data):
-            return False, (
-                "License not found. Please add a license to your model card using the `license` metadata or a"
-                " `license_name`/`license_link` pair."
-            )
-
-    # Enforce card content
-    if len(card.text) < 200:
-        return False, "Please add a description to your model card, it is too short."
-
-    return True, ""
-
-def is_model_on_hub(model_name: str, revision: str, token: str = None, trust_remote_code=False, test_tokenizer=False) -> tuple[bool, str]:
-    """Checks if the model model_name is on the hub, and whether it (and its tokenizer) can be loaded with AutoClasses."""
-    try:
-        config = AutoConfig.from_pretrained(model_name, revision=revision, trust_remote_code=trust_remote_code, token=token)
-        if test_tokenizer:
-            try:
-                tk = AutoTokenizer.from_pretrained(model_name, revision=revision, trust_remote_code=trust_remote_code, token=token)
-            except ValueError as e:
-                return (
-                    False,
-                    f"uses a tokenizer which is not in a transformers release: {e}",
-                    None
-                )
-            except Exception as e:
-                return (False, "'s tokenizer cannot be loaded. Is your tokenizer class in a stable transformers release, and correctly configured?", None)
-        return True, None, config
-
-    except ValueError:
-        return (
-            False,
-            "needs to be launched with `trust_remote_code=True`. For safety reason, we do not allow these models to be automatically submitted to the leaderboard.",
-            None
-        )
-
-    except Exception as e:
-        return False, "was not found on hub!", None
-
-
-def get_model_size(model_info: ModelInfo, precision: str):
-    """Gets the model size from the configuration, or the model name if the configuration does not contain the information."""
-    try:
-        model_size = round(model_info.safetensors["total"] / 1e9, 3)
-    except (AttributeError, TypeError):
-        return 0  # Unknown model sizes are indicated as 0, see NUMERIC_INTERVALS in app.py
-
-    size_factor = 8 if (precision == "GPTQ" or "gptq" in model_info.modelId.lower()) else 1
-    model_size = size_factor * model_size
-    return model_size
-
-def get_model_arch(model_info: ModelInfo):
-    """Gets the model architecture from the configuration"""
-    return model_info.config.get("architectures", "Unknown")
-
-def already_submitted_models(requested_models_dir: str) -> set[str]:
-    """Gather a list of already submitted models to avoid duplicates"""
-    depth = 1
-    file_names = []
-    users_to_submission_dates = defaultdict(list)
-
-    for root, _, files in os.walk(requested_models_dir):
-        current_depth = root.count(os.sep) - requested_models_dir.count(os.sep)
-        if current_depth == depth:
-            for file in files:
-                if not file.endswith(".json"):
-                    continue
-                with open(os.path.join(root, file), "r") as f:
-                    info = json.load(f)
-                    file_names.append(f"{info['model']}_{info['revision']}_{info['precision']}")
-
-                    # Select organisation
-                    if info["model"].count("/") == 0 or "submitted_time" not in info:
-                        continue
-                    organisation, _ = info["model"].split("/")
-                    users_to_submission_dates[organisation].append(info["submitted_time"])
-
-    return set(file_names), users_to_submission_dates

src/submission/submit.py

@@ -1,119 +0,0 @@
-import json
-import os
-from datetime import datetime, timezone
-
-from src.display.formatting import styled_error, styled_message, styled_warning
-from src.envs import API, EVAL_REQUESTS_PATH, TOKEN, QUEUE_REPO
-from src.submission.check_validity import (
-    already_submitted_models,
-    check_model_card,
-    get_model_size,
-    is_model_on_hub,
-)
-
-REQUESTED_MODELS = None
-USERS_TO_SUBMISSION_DATES = None
-
-def add_new_eval(
-    model: str,
-    base_model: str,
-    revision: str,
-    precision: str,
-    weight_type: str,
-    model_type: str,
-):
-    global REQUESTED_MODELS
-    global USERS_TO_SUBMISSION_DATES
-    if not REQUESTED_MODELS:
-        REQUESTED_MODELS, USERS_TO_SUBMISSION_DATES = already_submitted_models(EVAL_REQUESTS_PATH)
-
-    user_name = ""
-    model_path = model
-    if "/" in model:
-        user_name = model.split("/")[0]
-        model_path = model.split("/")[1]
-
-    precision = precision.split(" ")[0]
-    current_time = datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%SZ")
-
-    if model_type is None or model_type == "":
-        return styled_error("Please select a model type.")
-
-    # Does the model actually exist?
-    if revision == "":
-        revision = "main"
-
-    # Is the model on the hub?
-    if weight_type in ["Delta", "Adapter"]:
-        base_model_on_hub, error, _ = is_model_on_hub(model_name=base_model, revision=revision, token=TOKEN, test_tokenizer=True)
-        if not base_model_on_hub:
-            return styled_error(f'Base model "{base_model}" {error}')
-
-    if not weight_type == "Adapter":
-        model_on_hub, error, _ = is_model_on_hub(model_name=model, revision=revision, token=TOKEN, test_tokenizer=True)
-        if not model_on_hub:
-            return styled_error(f'Model "{model}" {error}')
-
-    # Is the model info correctly filled?
-    try:
-        model_info = API.model_info(repo_id=model, revision=revision)
-    except Exception:
-        return styled_error("Could not get your model information. Please fill it up properly.")
-
-    model_size = get_model_size(model_info=model_info, precision=precision)
-
-    # Were the model card and license filled?
-    try:
-        license = model_info.cardData["license"]
-    except Exception:
-        return styled_error("Please select a license for your model")
-
-    modelcard_OK, error_msg = check_model_card(model)
-    if not modelcard_OK:
-        return styled_error(error_msg)
-
-    # Seems good, creating the eval
-    print("Adding new eval")
-
-    eval_entry = {
-        "model": model,
-        "base_model": base_model,
-        "revision": revision,
-        "precision": precision,
-        "weight_type": weight_type,
-        "status": "PENDING",
-        "submitted_time": current_time,
-        "model_type": model_type,
-        "likes": model_info.likes,
-        "params": model_size,
-        "license": license,
-        "private": False,
-    }
-
-    # Check for duplicate submission
-    if f"{model}_{revision}_{precision}" in REQUESTED_MODELS:
-        return styled_warning("This model has been already submitted.")
-
-    print("Creating eval file")
-    OUT_DIR = f"{EVAL_REQUESTS_PATH}/{user_name}"
-    os.makedirs(OUT_DIR, exist_ok=True)
-    out_path = f"{OUT_DIR}/{model_path}_eval_request_False_{precision}_{weight_type}.json"
-
-    with open(out_path, "w") as f:
-        f.write(json.dumps(eval_entry))
-
-    print("Uploading eval file")
-    API.upload_file(
-        path_or_fileobj=out_path,
-        path_in_repo=out_path.split("eval-queue/")[1],
-        repo_id=QUEUE_REPO,
-        repo_type="dataset",
-        commit_message=f"Add {model} to eval queue",
-    )
-
-    # Remove the local file
-    os.remove(out_path)
-
-    return styled_message(
-        "Your request has been submitted to the evaluation queue!\nPlease wait for up to an hour for the model to show in the PENDING list."
-    )

text_content.py

@@ -0,0 +1,41 @@
+HEAD_TEXT = """
+This is the official leaderboard for 🏅StructEval benchmark. Starting from an atomic test objective, StructEval deepens and broadens the evaluation by conducting a **structured assessment across multiple cognitive levels and critical concepts**, and therefore offers a comprehensive, robust and consistent evaluation for LLMs. 
+
+Please refer to 🐱[StructEval repository](https://github.com/c-box/StructEval) for model evaluation and 📖[our paper]() for experimental analysis.
+
+🚀 **_Latest News_** 
+* [2024.8.2] We released the first version of StructEval leaderboard, which includes 21 open-sourced language models, more datasets and models as comming soon🔥🔥🔥.
+
+* [2024.7.31] We regenerated the StructEval Benchmark based on the latest [Wikipedia](https://www.wikipedia.org/) pages (20240601) using [GPT-4o-mini](https://openai.com/index/gpt-4o-mini-advancing-cost-efficient-intelligence/) model, which could minimize the impact of data contamination🔥🔥🔥.
+"""
+
+ABOUT_TEXT = """# What is StructEval?
+Evaluation is the baton for the development of large language models.
+Current evaluations typically employ *a single-item assessment paradigm* for each atomic test objective, which struggles to discern whether a model genuinely possesses the required capabilities or merely memorizes/guesses the answers to specific questions.
+To this end, we propose a novel evaluation framework referred to as ***StructEval***. 
+Starting from an atomic test objective, StructEval deepens and broadens the evaluation by conducting a **structured assessment across multiple cognitive levels and critical concepts**, and therefore offers a comprehensive, robust and consistent evaluation for LLMs.
+Experiments demonstrate that **StructEval serves as a reliable tool for resisting the risk of data contamination and reducing the interference of potential biases**, thereby providing more reliable and consistent conclusions regarding model capabilities. 
+Our framework also sheds light on the design of future principled and trustworthy LLM evaluation protocols.
+
+# How to evaluate?
+Our 🐱[repo](https://github.com/c-box/StructEval) provides easy-to-use scripts for both evaluating LLMs on existing StructEval benchmarks and generating new benchmarks based on StructEval framework.
+
+# Contact
+If you have any questions, feel free to reach out to us at [[email protected]](mailto:[email protected]).
+"""
+
+CITATION_BUTTON_LABEL = "Copy the following snippet to cite these results"
+
+CITATION_BUTTON_TEXT = r"""
+comming soon
+"""
+
+ACKNOWLEDGEMENT_TEXT = """
+Inspired from the [🤗 Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard).
+"""
+
+
+NOTES_TEXT = """
+* On most models on base MMLU, we collected the results for their official technical report. For the models that have not been reported, we use opencompass for evaluation.
+* For other 2 base benchmarks and all 3 structured benchmarks: for chat models, we evaluate them under 0-shot setting; for completion model, we evaluate them under 0-shot setting with ppl.
+"""

utils.py

@@ -0,0 +1,32 @@
+from dataclasses import dataclass
+
+
+@dataclass
+class ColumnContent:
+    name: str
+    type: str
+    displayed_by_default: bool
+    hidden: bool = False
+
+
+def fields(raw_class):
+    return [
+        v for k, v in raw_class.__dict__.items() if k[:2] != "__" and k[-2:] != "__"
+    ]
+
+
+@dataclass(frozen=True)
+class AutoEvalColumn:  # Auto evals column
+    model = ColumnContent("model", "markdown", True)
+    base_average = ColumnContent("BaseAverage", "number", True)
+    struct_average = ColumnContent("📝StructAverage", "number", True)
+    
+    base_mmlu = ColumnContent("BaseMMLU", "number", True)
+    struct_mmlu = ColumnContent("📝StructMMLU", "number", True)
+    
+    base_arc = ColumnContent("BaseARC", "number", True)
+    struct_arc = ColumnContent("📝StructARC", "number", True)
+    
+    base_openbook = ColumnContent("BaseOpenbook", "number", True)
+    struct_openbook = ColumnContent("📝StructOpenbook", "number", True)
+